Procedures for obtaining consent and ensuring confidentiality were proposed by the ethical research committees in The Sudan. Written permission to conduct the study was thus obtained from the authorities at the Ministry of Health and Ministry of Education, locality
administration and individual school administration. Verbal informed consent was obtained from the participants.
4 RESULTS 4.1 Sample profile
A total of 1109 (response rate 99%) (49.9% boys, 50.2% public school attendees) from 58 schools completed the interview followed by the clinical examination.
Paper I: Oral health status of 12-year-old schoolchildren in Khartoum State, The Sudan; a school-based survey
The mean DMFT for 12-year-olds was 0.42. Private school attendees had significantly higher DMFT (0.57) than public school attendees (0.40). The untreated caries prevalence (deciduous and permanent teeth) was 30.5%. Caries experience (DMFT > 0) affected 24% of the schoolchildren. Furthermore, the SiC was 1.4 for the adjusted sample and 1.6 and 1.4 for private and public school attendees respectively. Forty eight percent of the examined schoolchildren visited a dentist for treatment, and of those 96% visited because they suffered toothache or trauma. There was no statistically significant association between reason for visit and caries experience. The mean GI for the six index teeth was 1.05 (CI 1.03 – 1.07) and the mean PI was 1.30 (CI 1.22 – 1.38). . The prevalence of fluorosis was low (11.9%).
Fluorosis and oral hygiene indices were not associated with any of the non biological determinants and caries experience. Multivariate analysis disclosed that caries experience (DMFT > 0) was significantly and directly associated with socioeconomic status (IRR 1.23 (95% CI 1.02-1.47)). Private school attendees were associated with a higher mean DMFT.
Almost one third (30.3%) of the private school attendees experienced dental caries versus 23.5% of the public school attendees (details in Paper 1).
Paper II: Correlation between caries experience in Sudanese schoolchildren and dietary habits according to a food frequency questionnaire and a modified 24-hr recall method.
BMI calculations showed the representative prevalence as follows: underweight schoolchildren (35.8%), healthy children (55.8%), those at risk of becoming overweight (5.2%) and those overweight (3.5%). The following variables were tested individually with the independent outcomes, and only those that showed a statistically significant association were inserted in the multiple variable ANOVA model: SES, school sector, gender, parental education, locality, caries experience and BMI. Using sum FFQ score (mean 20.0, SD 3.5) as the dependent variable and SES, parental education and locality as fixed factors disclosed that locality (F=3.2, d.f.= 6, p=0.004) and SES (F=4.0, d.f. = 1, p=0.46) were statistically significant with no statistically significant interaction between the variables. However, the effect size was small (<0.02) for both. Post-hoc comparisons using the Tukey HSD test indicated higher mean consumption in the Sharq Elnil district than in Khartoum and Ombada, and higher mean consumption in Bahri than in Khartoum.
No significant association was found between the total frequency of intake of sugar-containing items reported on the FBC questionnaire and caries experience (DMFT >0:
n=298). The seven food items were run in a multiple variable logistic regression model (Nagelkerke 0.026) alongside socio-demographic variables, with the Higher Caries Experience Group (HCEG) as the dependent. This disclosed a statistically significant association for consumption of soft drink (OR 1.5 95% CI (1.0 – 2.4)).
Paper III: Caries experience and quantification of Streptococcus mutans and Streptococcus sobrinus in saliva of Sudanese schoolchildren
The mean ratio of fold differences of S. mutans to S. sobrinus was 0.77 (sd 5.4) and 2.29 (sd 6.0) for samples obtained from caries-free and caries-active individuals, respectively. This
suggested a higher proportion of S. sobrinus than S. mutans in the caries-active group than in the caries-free group. It was 3 times more likely to detect both bacteria in individuals with caries activity over their counterparts. Furthermore, it was three times less likely to detect S.
sobrinus alongside S. mutans in caries free individuals. There were no significant associations found between children with caries active lesions and their oral hygiene expressed in terms of GI, PI and tooth brushing frequency. An association was found between children with caries active lesions and the frequent consumption of sticky dessert and higher SES.
Paper IV. Evaluation of oral health-related quality of life among Sudanese schoolchildren using Child-OIDP inventory
The instrument showed acceptable psychometric properties and is considered as a valid, reliable (Cronbach’s alpha 0.73) and practical inventory for use in this population. An impact was reported by 54.6% of the schoolchildren. The highest impact was reported on eating (35.5%) followed by cleaning (28.3%) and the lowest impacts were on speaking (8.6%) and social contact (8.7%). Problems which contributed to all eight impacts were toothache, sensitive teeth, exfoliating teeth, swollen gums and bad breath. Toothache was the most frequently associated cause of almost all impacts in both private and public school attendees.
After adjusting for confounders in the 3 multiple variable regression models (whole sample, public and private school attendees), active caries maintained a significant association with the whole sample (OR 2.0 95% CI 1.4-2.6) and public school attendees (OR 3.5 95% CI 2.1-5.6), and higher SES was associated with only public school attendees’ Child-OIDP (OR 1.9 95% 1.1-3.1).
5 DISCUSSION
This section deals with important methodological issues and considers the main findings of the papers constituting the present thesis. Psychometric evaluation of Child-OIDP is described in detail in Paper IV.
5.1 Methodological aspects of the study
Data for the studies were collected using standard survey methods under field conditions, providing estimates of variables for a specific target population (Moser and Kalton, 1971).
The questionnaires included close-ended questions with several answers to improve the accuracy of responses. The main strength of this approach and one of the advantages of a sample survey approach is that it yields information on several variables in a large number of subjects, at relatively low cost. As discussed in the separate papers, this design is however, subject to errors with respect to sampling, non-coverage, non-response and measurement (Locker, 2000). The major sources of error are highlighted in the following section.
The interview as a whole, including the clinical examination, took approximately 20 minutes.
The variety of questions probably eliminated questionnaire fatigue.
5.1.1 Reliability
Test-retest: Several measures were taken to ensure acceptable data quality. For this, pilot studies were conducted and field assistants were trained on how to conduct the interviews.
Reliability Cohen’s Kappa was applied for test-retest reliability to examine measurement bias.
Test-retest reliability (reproducibility) refers to the extent of measurement consistency between different points in time. The interpretation of the Kappa values were as follows: <0 = no agreement, 0.0-0.2 = slight agreement, 0.21-0.40 = fair agreement, 0.41-0.60 = moderate
agreement, 0.61-0.80 = substantial agreement, 0.81-1.00 = almost perfect agreement (Landis and Koch, 1977). With regard to clinical examination (one trained dentist), the Kappa value for DMFT was 0.83 for 45 schoolchildren re-examined within 14 days, indicating almost perfect agreement. Reproducibility for GI and PI was not tested because all children were given oral hygiene instruction after the clinical examination and it was expected that in response, GI and PI scores would have improved at re-examination 14 days later. In this case, it would be difficult to distinguish between observed differences due to lack of consistency or due to true changes of the characteristics.
The principal questionnaire on socio-demographic and behavioural characteristics, the food frequency questionnaire and the Child-OIDP were reintroduced to a sample of 20 randomly selected 12-year-old schoolchildren. For logistical reasons, the test–retest interval for the questionnaires was 10 days. Kappa values for test-retest of the principal questionnaire ranged from 0.55 (knowledge) to 0.97 (wealth index). The data gathered by the questionnaire can improve in the 10-day interval between the reproducibility tests, thus these finding were accepted. The other Kappa values were of acceptable agreement.
Reproducibility test results of the FFQ showed that the Kappa-values for the sugar-sweetened items ranged from -0.03 to 0.89 (sugar-sweetened hot beverage, chocolates). Based on the Kappa results, the sugar sweetened hot beverage report was excluded from the analysis.
With respect to the Child-OIDP questionnaire, weighted Cohen’s Kappa was 0.70 for eating.
The Kappa value was 1.00 for the following Child-OIDP items; speaking, cleaning teeth, relaxing, sleeping, smiling, social contact and emotional state. These values were in moderate to substantial agreement according to Landis and Koch (Landis and Koch, 1977).
Measurement bias was thus minimal in the Child OIDP questionnaire.
Internal consistency denotes the interrelation of items in a scale and Cronbach’s alpha was calculated for this purpose. This test was conducted on FFQ, FBC and Child-OIDP questionnaires. Alpha coefficients above 0.80 are rated as exemplary, from 0.70 to 0.79 as extensive, and those in the range 0.60 – 0.69 indicate only moderate internal consistency (McDowell, 2006). The value for the FFQ and FBC (n=1109) was 0.50 for both instruments.
In this context the items are independent of each other, suggesting that no similar scores should be expected. Thus, a score of 0.50 was acceptable for the FFQ and FBC.
For Child-OIDP the standardized Cronbach’s alpha coefficient was 0.73 for the entire sample: 0.78 for public and 0.67 for private school attendees, all within the range of moderate to extensive consistency.
5.1.2 Validity
Validity is the ability of a tool to measure what it is intended to measure (McDowell, 2006).
Internal validity implies validity of inference for the source population of study subjects.
Several types of bias can affect the internal reliability (Sackett, 1979).
Diagnostic bias: Dental caries is best diagnosed in a dental surgery using visual, tactile and radiographic records. However, following the WHO standardized recording criteria in the field and under natural sunlight, only frank dentine caries could be diagnosed. This, however, enabled comparison of findings with other studies using the same index. Caries experience is thus under-estimated in all (Paper I).
Information bias: Face validity is the validity of a test at face value, and it reflects the immediate understanding of the questions. No major changes in the questionnaire were necessary after the pilot study, indicating good face validity. The interviewer confirmed that all the children understood the questions properly before answering.
Subjective oral health conditions were assessed using self-report methods, which are prone to recall and information/social desirability bias. Socially desirable answers were likely
controlled for by interviewing each student individually, away from their class teachers and colleagues. The dentist interviewer, although a stranger to the child, could unintentionally have influenced the children to provide socially desirable answers.
Confounding: Potentially confounding variables were controlled for through stratification and in multiple variable models. These limitations may have compromised the internal validity of the study.
External validity is the validity of generalized inferences in scientific studies. In this study, sample calculation and sampling procedures were optimised to ensure that the results of this study could be generalized to all 12-year-old schoolchildren in Khartoum State, thus minimizing selection bias. Paper I presents both the results of analysis of the sample with equal numbers of schoolchildren from the private and public sectors (1:1) and adjusted values according to the population of schoolchildren (public/private ~ 7:1).
Sampling error: The cluster sampling design employed provided advantages of simplicity and cost effectiveness and practicability in developing regions where school registers were incomplete. Furthermore, all estimates were adjusted for at the analysis stage using Stata. The cluster sizes were small, averaging 20 students per school. In all, 58 clusters were included.
After statistical analysis, the point estimates were the same before and after adjusting for cluster sampling. The only changes observed where in the slight narrowing of the confidence interval after adjustment.
The number of saliva samples tested was not calculated to yield results generalisable to the target population, but to test the hypothesis that an association existed between salivary parameters and the caries and the non-caries group.
Effect on non-response error: Adequacy of response rates may be rated as good (more than 80%), acceptable (70 – 79 %), suspect (55 - 69%), and unacceptable (less than 55% as (Locker, 2000). The response rate in this study was good (99%), giving further strength to the external validity of the study. A few children (fewer than 10, mostly girls) opted not to give saliva samples. This was understandably because of the procedure of spitting into a cup. No information was obtained from the non-participants. Although non-response is not a random process, it is unlikely that non-response error, which is a function of response rate and the magnitude of the difference between responders and non-responders, has biased the results of the present study.
5.1.3 Cross cultural adaptation
Both the dietary habit questionnaires were based on the results of five in-depth interviews, thus minimizing the likelihood that the questionnaires were culturally inappropriate.
The original Child OIDP questionnaire was an English language version. It was translated to and from Arabic and initially tested for cultural adaptability in a group of 12-year-old schoolchildren. Ideally a questionnaire should have been designed specifically for the Sudanese child population, in the local language/dialect. However, with the limited resources available it was more feasible to test a pre-designed questionnaire. The pilot study and discussion with colleagues demonstrated that this questionnaire was culturally acceptable.
5.2 Comments on the main findings
Dental caries and its sequelae have been the main concern of most dentists worldwide. This study reported results that may be extrapolated to all 12-year-old children living in the State of Khartoum. This state was specifically selected for study because it is the most
cosmopolitan and most densely populated in the country and includes various ethnic groups.
People migrate to Khartoum not only because as the capital it offers better living opportunities, but also because of internal displacement as a result of war and drought.
The index age of 12years was adopted in order to enable collation with previous studies in The Sudan and elsewhere.
5.2.1 Oral health status (Papers I and IV)
Dental caries prevalence and severity in terms of DMFT was very low (0.4). The impression gained from previous studies of oral health in The Sudan, was a trend towards increasing caries prevalence as described in the introduction (section 1). In contrast, the findings disclose a decline in caries experience among Sudanese 12-year-old schoolchildren. DMFT prevalence among the 12-year-olds of countries neighbouring The Sudan, reported after the year 2000, show that current DMFT values in Khartoum State are in the low range, comparable with the African populations in Tanzania and Nigeria (Lagos) where DMFT values of 0.3 and 0.46 (Agbelusi and Jeboda, 2006) were reported among 12-year-old children in 2004 and 2003/04 respectively. However, culturally similar neighbouring countries report higher values: Saudi Arabia reported a DMFT of 5.9 in 2002 (WHO Oral Health Country/Area Profile Programme) and the United Arab Emirates a DMFT of 1.6 in 2009 (El-Nadeef et al., 2009). The global caries burden was calculated to 1.61 DMFT for the year 2004, representing a continuous decline in most recent years (Bratthall, 2005). The condition in Khartoum State, in terms of current mean DMFT, is encouraging. However, despite the improvement, caries is unevenly distributed in the child population; a small proportion of individuals carry most of the disease burden. In addition, the proportion of teeth with active caries that went untreated (D
component of DMFT) has improved only slightly from the last report in 1986: from 98% of 12-year-olds, to 90% in the present study (Ibrahim et al., 1986).
Furthermore, the burden of disease was measured not only by the above clinical tools, but also subjectively. This report provides new and detailed evidence of the Child-OIDP of public and private school attendees in Khartoum State (details in Paper IV). The prevalence of oral OIDP in 12-year-old schoolchildren in Khartoum State (54.6%) was almost twice that reported in a similar age group in Tanzania (28.6%)(Mtaya et al., 2007). With the exception of the UK, all the remaining countries had higher impact prevalence, highlighting socio-cultural variations in the Child-OIDP (Yusuf et al., 2006, Tubert-Jeannin et al., 2005, Bernabe et al., 2008, Cortes Martinicorena et al., 2009, Castro et al., 2008, Bianco et al., 2009). Despite the high prevalence of impact on daily performance, the severity of impact was rarely high: most reports were of little or moderate intensity, and private school attendees reported a higher frequency of severe and very severe intensities compared to their public school counterparts.
Despite the low prevalence of dental caries experience (24%), a significant association was found with an average moderate intensity of Child-OIDP. The finding that more children reported a subjective impact caused by toothache (55%) than indicated by objectively assessed prevalence of caries experience (24%) suggested that clinical examination might underestimate treatment need.
5.2.2 General risk assessment based on the risk factor model in this population (based on Figure 3)
An attempt is made to analyse the caries risk in this population using the model illustrated in Figure 3. Not all the variables in this model were investigated in this study. Figure 10 illustrates the model with the examined variables.
Figure 10. Examined variables based on the Peterson model.
Utilization of dental services in this population was evaluated in terms of dental attendance habits: 52% had never previously been examined by a dentist. The reasons for going to the dentist and the care index (F component of DMFT) are similar to those in developing countries (Petersen, 2003).
Environmental risk factor, represented by nutritional status, was expressed in terms of BMI, which may be considered as a marker for nutritional status in some cultures. Although the water fluoride level was not known, this was estimated indirectly by recording the severity of fluorosis. Under socio-cultural risks, an association was disclosed between caries experience and children of higher SES. The indicators of socioeconomic status (SES) used in the present study were education and lifestyle indicators (household assets, house ownership, dwelling structure materials, family size and house size). Risk behaviour was evaluated in terms of brushing frequency and intake of sugary snacks. However, the type of toothpaste was not recorded. The individual contributions of these variables are discussed in Papers I, II and IV.
The combined contribution of the variables was examined in regression models (Paper I, II, IV) and the main findings are discussed below as indicators of caries experience and impact on daily performance.
Outcome:
Prevalence of dental caries Odds Ratio/prevalence ratio Caries severity
Oral health related quality of life (Child-OIDP)
1. Eating food 2. Speaking clearly 3. Cleaning your mouth 4. Relaxing 5 Emotional state
6. Smiling, laughing and showing your teeth 7. Carrying out your schoolwork) 8. Contact with people Use of oral health services:
History of dentist visit
Nutrition status Biological determinants:
Salivary mutans streptococci levels
5.2.3 Indicators of poor oral health (Papers I, II, III, IV)
Caries experience was associated with the private school sector, higher SES and the prevalence of S. sobrinus in saliva. Frequent consumption of soft drinks was associated with the higher caries group. Child-OIDP was associated with active caries in all the children and a higher SES in public school attendees. These findings further suggest the need for subjective assessment of oral health to complement conventional objective clinical assessment.
Recently there has been an increased concern about social gradients of dental caries, because oral health inequality has been increasing, even in countries with well-developed health care systems (Watt, 2007). In the present studies, schoolchildren of higher socioeconomic status were more likely to experience dental caries and public school attendees with a higher SES more likely to report an oral health related impact on their daily performance. This is in contrast with a previous report from The Sudan, over two decades ago (Ibrahim et al., 1986) and does not support the general belief that lower socioeconomic status is usually associated with caries experience and disease as a whole. In a review of the epidemiology of SES and health, Marmot concluded that the causes of inequalities in health within countries are similar to those between countries and emphasized the association between SES and health (Marmot, 1999). A literature review on social class and dental caries found higher caries experience in lower social classes, although five studies did not find any difference and two reported lower
Recently there has been an increased concern about social gradients of dental caries, because oral health inequality has been increasing, even in countries with well-developed health care systems (Watt, 2007). In the present studies, schoolchildren of higher socioeconomic status were more likely to experience dental caries and public school attendees with a higher SES more likely to report an oral health related impact on their daily performance. This is in contrast with a previous report from The Sudan, over two decades ago (Ibrahim et al., 1986) and does not support the general belief that lower socioeconomic status is usually associated with caries experience and disease as a whole. In a review of the epidemiology of SES and health, Marmot concluded that the causes of inequalities in health within countries are similar to those between countries and emphasized the association between SES and health (Marmot, 1999). A literature review on social class and dental caries found higher caries experience in lower social classes, although five studies did not find any difference and two reported lower